Date: Thu, 24 Apr 1997 11:45:04 -0700 (PDT) From: Simon Shapiro <Shimon@i-Connect.Net> To: Andreas Klemm <andreas@klemm.gtn.com> Cc: freebsd-bugs@freebsd.org, freebsd-hackers.freebsd.org@sendero.i-connect.net Subject: Re: RELELNG_2_2 is Broken - Help!!! Message-ID: <XFMail.970427164720.Shimon@i-Connect.Net> In-Reply-To: <19970424142306.59692@klemm.gtn.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi Andreas Klemm; On 24-Apr-97 you wrote:
> On Wed, Apr 23, 1997 at 05:21:38PM -0700, Simon Shapiro wrote:
> [-- Warning: iso-8859-8 is not compatible with your display.]
>
> > What is the value of PCI_SUBID_REG2 ??
>
> In the commit message to pci.c Stefan wrote 0x40. So one had a
> nice chance to include the define into pcireg.h ;-)
Thanx! That will help. I am running CVSup several times a day so my next
build should catch it.
...
> I'm running 2.2-STABLE and have zero problems at all ! Runs perfectly
> stable ! Maybe you should check your kernel config file or upgrade
> to 2.2-STABLE and do a make world ???
There is nothing wrong with the kernel config file. The same as on BETA.
Your configuration is probably different than mine. There is an
acknowledged set of problems, that later releases solve;
ahc problems, scsi problems, etc. Besides, I may be dense and slow, but I
cannot get cvs to checkout for me any release other than current or
RELENG_2_2 without missing files, etc.
>
> > Alternative: How do I checkout of CVS /usr/src/sys as of 13-Feb-97?
> It is
> > the last kernel I know of to exhibit usable level of stability.
>
> With the -D option, which allows you to checkout a release branch
> or HEAD (-current) at a certain date ...
Except that is misses files this way (sys/lock.h and kern/kern_lock.c,
for example. does little to encourage me to trust it :-(
...
> Just in case that you read the mailing lists and news you might have
> noticed, that Justin worked very hard on that and if you update your
> machine to 2.2-STABLE, you'll see, that all the problems will go
> away. I have a really rock solid 2.2-STABLE here, no matter if I
> do large I/O disk benchmarks, dumps and such !
300 disk drives? P6? FDDI? Bash? I can reproduce the panic at will.
It is nasty as it takes an hour or so of fsck to re-boot, but it panics,
very specifically with RELENG_2_2. I re-build the system (CVSup,
cvs update, make world) twice a day. No difference. I will replace the
RAM on the machine after typing this note, but doubt it very much.
> I'll give you a kernel config file as reference in the attachement.
Thanx!
> > I am trying to get a 2.2 kernel that runs for more than 5 minutes
> without a
> > panic on the open problem (with short shell scripts, etc.).
>
> _The_ open problem ??? Is it a well known problem ? ;-)) Well, I
> have _zero_ problems in this area. FreeBSD acts as a Internet gateway
> here and I' gettings nearly all mailing-lists, have squid running,
> a WWW Server ... and no problems with open.
I belive FreeBSD is stable and wonderful. I commited a multi million dollar
project to this O/S in leu of Slowlaris, after all. I just need some help.
do you knwon of anyone who is willing to help? I can pay pretty well for
short/long/permanent contract/employment.
> I suggest you tell your problems more verbously.
I did before, and here it is again:
a. Boot any RELENG_2_2 release.
b. Start X11; In .xinit put xfm
c. Click on a ``Mail'' button; This will run a one/two line shell script
d. Observe the system freeze, completely.
e. If you are lucky/patient, quickly, before it dies, do ctl-alt-F1 and
you will see:
a panic, which, when traced, will show:
Fatal trap 12: Page fault at 0xf71e0014
_spec_open + 0x6e
_spec_open
_vm_open
_open
_syscall
__Xsyscall (bash)
> But anyway ... there is a STABLE tree, that's RELENG_2_2,
> either you cvsup the sources or the cvs repository, if the latter
> is the case for you, simply checkout RELENG_2_2
>
> # cd /somewhere_with_about_150_MB_free_space
> # cvs checkout -r RELENG_2_2 src
> # cd src && time make world >& world.log &
I am doing EXACTLY that. No difference. make world completes without a
hitch. but the problems above are very real.
> Since you have problems with your SCSI subsystem it's surely a
> good idea to build and install a new stable kernel first before
> doing a make world.
Obviously. thanx. The SCSI problems are now very well isolated and
unrelated to the above:
a. Iomega Jaz spinning down and not responding in time to a command.
the BETA_A kernel (for which i do not have sources anymore :-((
panics but you can continue (assuming you were on VC0 :-) current 2.2
kernel panic and do not recover.
b. The experimental DPT driver confuses scsi_attachdevs(). this is on
a separate system altogether; differnet kernel, etc.
Thanx for ANY help!
Simon
P.S. Any garbage following this line?
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?XFMail.970427164720.Shimon>